Tumor neoantigens, which are immunogenic peptides arising from somatic mutations, have emerged as important determinants of effective human anti-tumor immune responses. Given the biologic and therapeutic importance of this class of antigen, the next priority to address is to understand the mechanisms by which neoantigens elicit immune responses and by which tumor-host immunity co-evolve. In doing so, we can better identify therapeutic challenges and opportunities, especially given the increased clinical availability of novel immunologic agents. To undertake this challenge, we need to have the best possible approach for detecting tumor neoantigens, an accurate method to detect the diversity of T cells responding to them, and informative sample cohorts to study this important biologic problem. To improve our detection sensitivity of neoantigens, we will integrate computational and experimental approaches to identify neoantigens discoverable from transcriptome data (i.e. gene fusions and splicing alterations) (Aim 1). Until now, neoantigen studies have almost exclusively focused on alterations detected by DNA sequencing (i.e. arising from missense mutations); where mutation detection algorithms are more mature. We will further seek to improve the accuracy of HLA binding prediction algorithms, as this is a key feature of neoantigen discovery workflows. The accuracy of these tools directly relates to the numbers of binding peptides used to train them, but sufficient data have been available for only few common HLA alleles. By experimentally creating large datasets of binding peptides across diverse HLA molecules at high-throughput using high-performance mass spectrometry, we will systematically generate high-quality data so that the rules surrounding peptide cleavage, display and binding in association with HLA molecules can be evaluated, and thereby improve binding prediction (Aim 2). Our final goal is to understand the coevolution of the immune response and tumor cells, specifically chronic lymphocytic leukemia (CLL). Tumor and immune cell populations are composed of genetically-defined distinct subclones, both of which can evolve over time. Central to the tumor-host immune cell interaction are the tumor antigen and antigen-specific T cells. Since tumor neoantigens have been implicated as key immunologic targets, we will interrogate whether the number and characteristics of personal tumor neoantigens in patient CLL samples change over time with spontaneous disease progression or following immunologic therapy. In parallel, we will measure the quantity and quality of neoantigen-specific T cells in th host microenvironment in serial marrow and peripheral blood specimens by single cell analysis of TCR repertoire and transcriptional state of neoantigen-specific T cells over time (Aim 3). Driven by urgent clinical and biological questions, we propose that a deep analysis of CLL-immune co-evolution using advanced computational algorithms, high performance mass spectrometry and single cell technologies will advance our understanding of how to best harness tumor immunity for cancer treatment.